Instrumentation capable of rapidly identifying individual aerosol particles at high sampling rates does not currently exist despite numerous practical applications and immediate need. The range of uses for such devices is enormous, spanning both civilian and military requirements. Historically, there are probably two major reasons for the absence of such units from the sophisticated electro-optical and electronic hardware available today.
First, of course, is the fact that particle identification per se is usually associated with concepts of chemical and spectroscopic analysis; viz. collecting the particles, reacting them with chemical and/or physical (electricity, heat, etc.) probes, and then analyzing the resultant spectra or reaction products. The very small size of each sample whose identification might be required (i.e., a single aerosol particle) makes such chemical approaches very difficult from an operational point of view. In the area of bacterial detection alone, for example, huge Federal resources were expended in the 1960's to build units that might collect and separate such particles from their ambient surroundings, maintain their viability, culture them or tag them with species specific fluorescent stains, and (eventually) yield the required detection alarms. Such systems were cumbersome, expensive, unable to operate under real time conditions, and generally ineffective. Another similar area of aerosol analysis that absorbed vast amounts of Federal funds in the 1970's related to the relatively simple detection of airborne asbestos fibers. The Environmental Protection Agency opted for "conventional" measurements of such particles such as X-ray spectroscopy and electron microprobe devices. When the impracticality of such approaches eventually became obvious, the EPA funded an optically oriented instrumentation technique developed by Leeds and Northrup. Large fibers were detected with the L&N Microtrac units, but unambiguous identification and application to smaller fibers were not obtainable.
The second historical reason for the lack of suitable aerosol identification instrumentation related to the huge amount of data handling that such units would have to perform. Until the advent of the microprocessor and, more importantly, the recently developed array processors, even the mere booking associated with the identification of thousands of particles was impossible from the practical point of view. These latter obstacles no longer exist, of course, and the availability of inexpensive, highly compact data handling hardware has resulted in considerably more attention being devoted to the real time aerosol identification problem.
Given the availability of suitable electronics, optical techniques are attractive as an instrumentation basis for an aerosol analyser, though not enough attention has been devoted to their practical implementation. Several particle counters and so-called "sizers" based on optical scattering techniques have been developed and marketed commercially during the past decade. But claims regarding the capability of these instruments to size particles accurately have been greatly exaggerated; probably because such measurements from the point of view of classical optical scattering have been considered simple and straightforward. For particles below a few thousand nanometers in diameter and/or exhibiting shape and structural irregularity and/or absorptive properties, Cooke and Kerker pointed out the shortcomings of these "conventional" optical analysers in 1975. Their analyses have been generally overlooked or ignored as both Federal Agencies (notably NASA and NOAA) and commercial instrument manufacturers continued to use their resources for the refinement of instrumentation in which they earlier had made large investments.
The desire to characterize aerosol particles by instrument means has invariably required some form of measurement that could at least yield a particle size. Optical methods have always seemed attractive in this regard and many instruments have been developed that incorporated an optical means of one sort or another to provide a size estimate. As alluded to by Cooke and Kerker, these optically-based particle counter/sizers do not generally yield correct answers, especially for submicron particle sizes. Particles in this latter size range, unfortunately, are among the most important types as they remain airborn for great periods of time, are responsible for major sources of visibility obscuration, and include important particles for the propagation of biological and chemical threats. More sophisticated optical approaches have also been developed which purport to be able to size particles well into the submicron region, but for the most part, they too have some serious shortcomings which preclude such measurement. Surprisingly, these flaws have very little to do with instrument design, but arise because of some simple oversight related to the underlying theory of the interaction of light with matter. An interesting example of this is the particle sizing interferometer (PSI) which yields extremely valuable information about large particles, and especially fibers. Unfortunately, for smaller particles, the fundamental assumption on which basis the system operates is invalid, viz. ". . . the light which is scattered by the particle is proportional to the observed flux illuminating it . . . ." Ever since Lorenz developed his theory to explain the scattering of light from spherical particles, it has been known that the amount of light scattered by such particles is not proportional to the particle's size. Indeed, for certain sizes and refractive index compositions, the amount of light intercepted and scattered by a particle can be many times the amount incident on its physical area, and this is especially true in the submicron (or resonance) region.
Now it has long been reasoned that if one could measure all of the light scattering properties of a single particle, one could in concept deduce its structure (shape, size, refractive index). Such measurements performed at a single wavelength are usually called the Stoke's parameters, or more generally, the Mueller matrices as discussed, for example by Thompson, et al in Applied Optics. However, it should be pointed out that these matrix elements (referred to as 4 states of incident polarization) are measured in a plane. Of the 16 assorted DLS patterns so recorded, only 7 are independent, corresponding to the 8 values of the two complex orthogonal amplitudes of the incident and scattered electric fields less a common phase of each. Insofar as single particles are concerned so elaborate a measurement does not seem necessary to deduce all of its optical parameters. Indeed, a single state of incident polarization may well yield equivalent data, when scattering measurements are recorded over the surface of a sphere rather than restricted to a plane. Be that as it may, after all the measurements have been made (if they can be made at all), the most difficult problem still remains: How can we deduce the particle's structure (shape, size, refractive index) from these measurements? In this disclosure we shall present a method and instrumentation design that will achieve this "inversion". But first we must explain what this means.
For the case of an inhomogeneous aerosol particle exhibiting spherical symmetry, there is a quantum mechanical analogue whose study has provided us useful guidance in the delineation of various systems parameters that are incorporated into our invention. R. Newton, for example, has shown that, within certain limits, the form of the scattering potential (analogous here to the dielectric composition of the scattering particle) may be deduced if a single phase shift be measured at all incident energies (all wavelengths), or if all the phase shifts be determined at a single energy (wavelength). The measurement of the phase shift .delta..sub.l, would be achieved by measurement of the associated scattering amplitude exp (i.delta..sub.l) for each value of l yielding an effectively non-zero .delta..sub.l. Experimentally, of course, one usually measures the scattered intensity which is proportional to the square of the magnitude of the total scattering amplitude. Thus the measurement of scattered intensity at a single wavelength does not provide for deduction of the total amplitude phase, nor does it permit the separation of the individual phase shift amplitude contributions, let alone their individual phases. Nevertheless, we generally assume that "embedded" in the measured scattered intensity, for a given state of incident polarization, is all the important phase shift information from which may be deduced the scattering particle's structure. Can indeed such measurements yield unique structural information?
It has generally been assumed for some time that the dielectric structure and composition of a spherically-symmetric particle can be deduced from a measurement (including various measurements at different incident polarizations) of the scattered light intensity over a reasonably broad range of scattering solid angles. Since the direct deduction of the phase shifts, and the reconstruction therefrom of the scattering particle's structure, is impossible, the usual method has consisted of parameterizing a spherically symmetric model and adjusting these parameters to obtain a least squares' fit to the data. This method has been applied successfully by Wyatt, and Wyatt and Phillips, to a variety of particle types in the sense that the parameterized particle structure so-derived has been both physically reasonable and in agreement with values derived from other methods.
The validity of the aforementioned "inversion" technique has never been proven formally except for a very simple two-dimensional case. R. Mirales has shown that a least squares' fit of experimental differential light scattering data will yield a unique dielectric value and size for the case of an infinite homogeneous cylinder. Nevertheless, the technique, at least for spherically symmetric scatterers, has become quite standard, though still rigorously unproven. It should be pointed out, however, that there are certain types of dielectric structures that scattering measurements cannot "sense". These include certain non-scattering, or "invisible" particles in the Kerker sense as well as structures within regions through which the incident wave cannot penetrate. For example, any material inside of a perfect conducting region will not contribute to the scattered waves. Also, particles of very high refractive index (real or complex) will not permit much penetration of the incident wave and preclude, therefore, deduction of some internal dielectric features.
Despite the success of the simple least squares fitting technique and its utility for even complex particle structures, there are some obvious problems with the method. Consider a homogeneous, isotropic, spherical particle. Only two physical parameters completely describe this particle at a fixed wavelength of incident light: its radius R and refractive index n. Depending upon the particle size, however, even the simplest scattering measurement can yield a very detailed scattering pattern. For example, relief plots made by Pendorf of the scattering properties of a single particle of refractive index 1.33 for vertically polarized incident light showed enormous variations of scattered intensity with angle (differential light scattering) as a function of size. These plots showed, furthermore, that as the particle size increases, the angular scattering pattern becomes more complex. Because of this complexity, more detailed measurements of the light scattering properties of individual particles have been made, often yielding hundreds of data points for larger particles. Yet only two physical parameters R and n characterize the particle exactly. Why should it be necessary to make so many measurements to derive so few parameters?
The same question may also be raised concerning more complex homogeneous particle structures such as rods (3 physical parameters: length, radius, and refractive index), ellipsoids (4 physical parameters: 3 semi-axes and refractive index), simple flakes (4 parameters: thickness, length, width, refractive index) homogeneous absorbing spheres (3 parameters: radius and real and imaginary parts of refractive index), etc. The present invention describes a method and apparatus by which means particle characterization may be achieved with fewer measurements, and greater accuracy in shorter time periods than has heretofore been possible.